|
内容記述 |
The Linear IFMIF Prototype Accelerator (LIPAc) is designed to accelerate 125mA deuteron beams in continuous wave. The beam from an ECR ion source is accelerated by an RFQ up to 5MeV and guided to a beam dump, after a horizontal bend. Next year a superconductive LINAC will be installed between the RFQ and beam dump to boost the beam energy up to 9MeV. Recently we started to introduce advanced optimization algorithm for various tasks involving the commissioning and operation of the machine. The beam current extracted from the ECR source is optimized by regulating the strength of 2 coils of the ECR with Gaussian Process Bayesian optimization. The optimal settings are found automatically within few minutes. We plan to tune additional settings: 2 tuners, gas flow and intermediate electrodes. We also present the optics simulation optimization with the use of genetic algorithms. The goal is to optimize a section of the transport line to improve the measurement uncertainty of the beam energy spread with the slit/grid method in a dispersive region including the bend. The strengths of 5 quadrupoles are tuned to obtain the Pareto front describing the minimum beam size for the highest dispersion at the grid location, with the constraint that the beam size is small enough along the whole accelerator to have negligible losses. The relative uncertainty of the set of measurements is then minimized from 25% to 1%. |
